Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method of noise reduction for multi-dimensional input data comprising a plurality of data points spaced from one another in each of multiple dimensions, the method comprising: a) defining at least one control lattice for the input data, each of the at least one control lattice having a plurality of control points separated by a predetermined spacing; b) superposing each of the at least one control lattice on the input data; c) defining, for each of the plurality of data points in the input data, a number of control points of the plurality of control points of the at least one control lattice surrounding the data point; d) computing, for each control point, a value which is a function of values of neighbouring data points; e) determining, for each data point of the plurality of data points in the input data, an approximation function from at least values of neighbouring control points, wherein determining the approximation function comprises applying a weighting to each of the values of the neighbouring control points based on a comparison of a value of the data point in the input data and the computed value of the neighbouring control point; and f) using the approximation function to filter the input data to generate filtered output data.
2. The method according to claim 1 , wherein the weighting is 1 if the difference of an absolute value of the the data point in the input data and the computed value of the neighbouring control point is lower than a predetermined threshold.
3. The method according to claim 2 , wherein the weighting is 0 if the difference of an absolute value of the the data point in the input data and the computed value of the neighbouring control point is higher than the predetermined threshold.
4. The method according to claim 1 , wherein the approximation function comprises an N-dimensional approximation function and step c) comprises, for each data point, defining 4N control points surrounding the data point.
5. The method according to claim 4 , wherein the N-dimensional approximation filter comprises a set of uniform cubic B-spline functions.
6. The method according to claim 5 , wherein step e) further comprises defining the N-dimensional approximation function as a linear combination of the neighbouring control points and the set of uniform cubic B-spline functions for each dimension.
7. The method according to claim 5 , further comprising the steps of pre-computing and tabulating the set of uniform cubic B-spline functions.
8. The method according to claim 1 , wherein step d) comprises pre-computing and tabulating values for the control points.
9. The method according to claim 1 , wherein applying the weighting comprises multiplying the value of the neighbouring control point by the weighting.
10. The method according to claim 1 , wherein a distribution of the weightings applied to the values of the neighbouring control points comprises a Gaussian distribution.
11. The method according to claim 1 , wherein the filtered output data comprises a same number of dimensions as the input data.
12. An image processing system comprising: an imaging system configured to capture at least one image; and circuitry configured to perform the steps of the method according to claim 1 .
The image processing system is designed for capturing and analyzing images to extract specific information. The system includes an imaging system that captures at least one image, which may be a still image or a sequence of images. The captured images are processed by circuitry that performs a method to analyze the images. This method involves detecting and identifying objects or features within the images, such as shapes, patterns, or other visual elements. The circuitry may also apply image enhancement techniques to improve the quality of the captured images before analysis. Additionally, the system may include functionality to compare the detected objects or features against a database or predefined criteria to determine their relevance or significance. The system is particularly useful in applications where automated image analysis is required, such as surveillance, quality control, medical imaging, or autonomous navigation. The circuitry ensures efficient processing of the images, allowing for real-time or near-real-time analysis depending on the application. The system may also include user interfaces or output mechanisms to display or transmit the results of the image analysis.
13. The image processing system according to claim 12 , wherein said imaging system comprises a three-dimensional imaging device for capturing depth information, and the input data comprises a depth map.
14. The image processing system according to claim 12 , further comprising a display; wherein the circuitry is further configured to cause the display to display the captured at least one image.
15. A non-transitory computer-readable data storage medium containing computer-executable instructions that, when executed by at least one computer processor perform a method comprising: a) defining at least one control lattice for the input data, each of the at least one control lattice having a plurality of control points separated by a predetermined spacing; b) superposing each of that least one control lattice on the input data; c) defining, for each of the plurality of data points in the input data, a number of control points of the plurality of control points of the at least one control lattice surrounding the data point; d) computing, for each control point, a value which is a function of values of neighbouring data points; e) determining, for each data point of the plurality of data points in the input data, an approximation function from at least values of neighbouring control points, wherein determining the approximation function comprises applying a weighting to each of the values of the neighbouring control points based on a comparison of a value of the data point in the input data and the computed value of the neighbouring control point; and f) using the approximation function to filter the input data to generate filtered output data.
This invention relates to data filtering techniques, specifically a method for processing input data using control lattices to generate filtered output data. The problem addressed is improving data filtering by dynamically adjusting the influence of control points based on local data characteristics. The method involves defining at least one control lattice for the input data, where each lattice consists of multiple control points spaced at predetermined intervals. These lattices are superposed onto the input data. For each data point in the input data, the system identifies the surrounding control points from the lattice. Each control point's value is computed as a function of neighboring data points. For every data point, an approximation function is determined using values from neighboring control points, with weights applied based on comparisons between the data point's value and the computed control point values. This weighting ensures that control points with values closer to the data point's value have greater influence. Finally, the approximation function is used to filter the input data, producing the filtered output. This approach enhances filtering accuracy by adaptively adjusting the contribution of control points based on local data variations, making it suitable for applications requiring precise data processing, such as image or signal filtering.
Unknown
January 26, 2021
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.